Using magnetic resonance imaging-based planimetry, we measured cortical and cerebral (cortical and ventricular) atrophy in 26 patients with Alzheimer's disease (AD) (age, 72.2 +/- 7.0) according to NINCDS-ADRDA criteria and 22 control subjects (age, 71.5 +/- 5.4). AD patients exhibited greater cerebral atrophy (p < .05) than control subjects. Cerebral atrophy was significantly correlated with age (r = .72, p < .0005) in healthy volunteers but not in AD patients. In AD patients, age of onset was negatively correlated with the estimated rate of disease-attributed cerebral degeneration ([observed atrophy--atrophy in normal aging calculated from the regression equation derived from the control group]/[duration of illness]) (r = -.54, p < .005). Multiple regression with interaction analysis demonstrated that age, age of onset, and their interaction successfully explained cerebral (R2 = .51, p < .05) and cortical (R2 = .64, p < .05) atrophy in patients with probable AD. Age of onset may be a strong predictor of the rate of cerebral degeneration in AD, and our results suggest that controlling age and the age of onset is essential in the quantitative study of AD.